[00:59:56] anomie: Thanks for looking into the "Script error" issue. :-) [02:03:10] halp [02:03:14] Error deleting file: The file "mwstore://local-multiwrite/local-deleted/r/u/o/ruosi6mzdnkvv3vgq0cz24p4tlao35d.jpg" is in an inconsistent state within the internal storage backends [02:13:04] aw all the devs are asleep [02:13:09] maybe they'll see this error in the morning [02:22:23] hi, is there a template or other wiki markup to list section headings from a page? thanks. [02:22:44] (like __TOC__, but I want to add toc of page A to to page B) [02:29:52] gry: I doubt it [03:47:55] Hello!?? [03:48:17] How to use RINDS Drive? [03:48:49] Have people in !?? [03:50:12] boom: is this a question about Wikimedia websites? [03:50:17] * Jasper_Deng doesn't understand the question [06:28:46] So, getting those 403 errors that you probably hear about all the time. I fixed my user-agent to my own personal one with an email address in it and I changed from GET to POST, but it's not working. Am i IP blocked and can I lift the block either by waiting or contacting someone? [06:37:13] calais: I don't think I have heard about 403 errors [06:37:34] what is the text of the error message? [06:40:26] TimStarling: http://www.mediawiki.org/wiki/API:FAQ#do_I_get_HTTP_403_errors.3F [06:40:50] what's that got to do with it? [06:41:10] TimStarling: I am getting the generic "Our servers are currently experiencing a technical problem." -error that supposedly comes from having the wrong user-agent [06:41:42] so a squid error message? [06:41:51] But the page has a status code of 403 and status_message "forbidden" [06:42:20] what is your IP? [06:42:38] is it the same one you are using on IRC? [06:43:00] TimStarling: I sent it in a private message. But it shoul be the same [06:44:54] I'm pretty sure it's not blocked in squid [06:44:57] what user-agent? [06:46:56] it started out as +Drupal (default drupal user agent for the function drupal_http_request), but I changed it to one that I'll send by private msg, since it contains my email [06:48:49] works for me [06:52:59] Thanks TS, got it sorted. [09:39:27] andre__: can you share the command you used to get the error in https://bugzilla.wikimedia.org/show_bug.cgi?id=23326#c13 ? [09:40:02] I get a different output with ghostscript, it would be useful for the people at https://commons.wikimedia.org/wiki/Commons:Bots/Work_requests#Fix_some_invalid_PDFs [09:54:12] Nemo_bis: it was just "gs filename", but I also get different output on this machine which still has gs-9.05 (**** File did not complete the page properly and may be damaged.) [13:12:11] andre__: but I have 9.06 too and I'm on fedora 18 too :( , but still different output [13:12:22] Nemo_bis, how weird [13:13:46] I'll check better I guess [13:14:12] hmm or maybe there's poppler doing something weird? I don't remember dependencies [13:17:33] does anybody have a working (minimal) example for a Javascript-based API fileupload via action=upload? I can't get it working – [13:20:34] (jQuery based example: https://dpaste.de/KcGMH/) [13:27:52] d_jan: according to http://stackoverflow.com/questions/11046684/php-file-upload-using-jquery-post you can't... [13:29:22] without using another library anyway [13:30:01] Krenair: Using the File-Api it should actually be possible. [13:30:33] (I'd be fine with an HTML5-only solution for now) [13:38:43] d_jan: hmm... https://developer.mozilla.org/en-US/docs/Using_files_from_web_applications [17:40:39] is there anyone out there that can help with AbuseFilter? [17:41:01] we are having hard time optimizing our filters [17:42:24] Hi [17:42:29] Sure [17:43:15] can you look at http://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci ? [17:43:36] any filters in specific? [17:44:06] they all have 0 hits [17:44:19] I hae gone through all but decreasing the "number" of conditions did not help. Is there anything else we need to focus on? [17:44:54] in particular, no. 6 is begi [17:45:00] rillke: i see filters with hits... [17:45:03] .. is being used frequently [17:45:08] triggered, I mean [17:45:10] ah ok [17:45:18] legoktm: but not in the table [17:45:32] https://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci/6 <-- do the namespace check first, then check id [17:45:47] rillke: i see them in the table... [17:46:05] column Hit count is empty [17:46:09] for me [17:46:17] http://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci?uselang=en [17:46:44] ok, fine [17:46:51] Ah, I have to log in [17:46:59] but we are using 52% of the condition limit [17:47:05] exhausting it... [17:47:51] what is the correct way to improve a specific filter? [17:50:40] for example, no. 6 is using 622 conditions! [17:50:42] Son 2.630 eylemde, bu filtreyle eşleşen 64 (2,43%). Ortalamada, çalışma süresi 2,06ms, ve koşul sınırının 622 koşulunu tüketiyor. [17:50:59] I am having hard time understanding this [17:51:09] its because you just updated it [17:51:09] 622 conditions? [17:51:13] in a few minutes itll drop down [17:51:30] yes, but the total number of conditions used does not change at all [17:51:41] Son 2.630 eylemde, 1.367 (%51,98) filtre 1.000 koşul sınırına erişti, ve 150 (%5,70) eylem şuanda etkin bir filtreyle eşleşti. [17:51:50] 52%??? [17:52:13] holy crap. [17:52:36] I am not sure what we are doing wrong [17:54:12] are we supposed to decrease the number of conditions at all times? I may be getting the idea wrong, not sure [17:54:19] ok so 6 is the issue right now. [17:54:38] yes, 6 is eating it up [17:54:57] the bad news is that we CANNOT deactivate it [17:55:25] Reedy: missing a comma here on line 1159: https://gerrit.wikimedia.org/r/#/c/65602/2/wmf-config/InitialiseSettings.php [17:55:36] we are having difficulty getting "good" article starts from anonymous users, that is why the filter is there [17:55:45] I poked someone who's better at writing filter rules than me to take a look [17:56:24] thanks lego [17:57:21] Superyetkin: You are trying to avoid that stubs are created, right? [17:57:31] (with filter 6) [17:58:36] avoid? no [17:58:49] we are trying to save good edits, discard bad ones [17:59:11] you cannot allow a new article with 100 a's and 200 f's in it [17:59:27] aaaaaaaaaaaaaaaaaaafafafaaaaaaaaaaaaaaaaaaafafaaaaaaaaaaaaaafafafaaffaafafaf [17:59:37] ok? [18:00:14] but you're not checking for repeated letters but for the size [18:00:32] yes, I wonder why this filter is eating up resources [18:01:01] (the article size; a small article is called [[:en:Wikipedia:Stub]]) [18:01:11] yes, I know [18:01:16] I am an active sysop :) [18:02:40] Superyetkin: oh yeah one thing he noticed is that you have tag enabled on a bunch of filters that are warn+disallow. you cant tag edits that are disallowed [18:02:57] what do we need to do to prevent the condition limit exhausting the resources? we are using 52%! [18:03:42] lego, why not tag even when we disallow something? I think it works fine [18:03:53] because you cant tag [18:04:03] do you mean it is "unnecessary"? [18:04:08] you can only tag edits that go through, but since you're disallowing it doesnt work [18:04:11] Superyetkin:The tag is something that is visible after the edit summary [18:04:27] oh, got it [18:04:30] you are right [18:06:34] Is there a convenient place to get a list of all the language codes for the various language wikipedias? [18:09:13] ok rilike, I removed the tag options now [18:09:28] what is the real issue with #6? [18:09:31] It will be still in the abuselog [18:09:51] elplatt: https://en.wikipedia.org/wiki/Template:Wikipedias [18:10:03] Who wanted help with trwiki filters? [18:10:06] what about this? [18:10:08] Son 2.675 eylemde, 1.388 (%51,89) filtre 1.000 koşul sınırına erişti, ve 152 (%5,68) eylem şuanda etkin bir filtreyle eşleşti. [18:10:17] we are using 51% of the condition limit [18:10:23] I cannot believe that [18:10:33] 51% of edits are hitting the condition limit and stopping in the edit filter [18:10:36] harlot: Superyetkin [18:10:39] Try [18:10:41] elplatt: also https://noc.wikimedia.org/conf/highlight.php?file=wikipedia.dblist [18:10:45] !("confirmed" in user_groups) & [18:11:03] (old_size == 0) & [18:11:06] Thanks, legoktm that second one will work. [18:11:19] (article_namespace == 0) & [18:11:24] (new_size < 150) & [18:11:26] harlot, why setting old_size to zero? [18:11:36] If you only want to match new articles [18:11:43] no [18:11:45] or article_articleid [18:11:55] a blanked article has a size of zero as well [18:12:06] #6 is doing that exactly [18:12:28] Then replace contains_any with rlike [18:12:50] contains_any is really slow [18:12:56] really? [18:13:01] yes [18:13:12] ok, thanks for that [18:13:25] You also don't have the first condition [18:13:25] !("confirmed" in user_groups) & [18:13:28] what is the effect of adding that one? [18:13:37] You probably don't need to be checking all long-term editors. [18:13:50] anybody who is confirmed / autoconfirmed won't be checked [18:14:03] but are we not increasing the number of conditions by doing that? this is the point I may be missing... [18:14:19] The filter stops processing after *any* check fails [18:14:23] but they will be checked once, rşght? [18:14:47] hmm, ok [18:14:58] Nothing more will be processed in that filter if the first check fails, saving resources. [18:15:00] so, we need to use that one [18:15:13] we have "autoconfirmed", though [18:15:40] "confirmed" is in "autoconfirmed" [18:15:49] is in? [18:15:56] what about "user"? [18:16:08] the text "confirmed" appears in "autconfirmed" [18:16:21] yeah [18:16:51] "user" if you want it to match all registered accounts [18:17:10] But you probably want the filter to check new accounts [18:18:34] thanks harlot [18:18:39] how about this one? [18:18:42] ****************************************** [18:18:43] !("autoconfirmed" in user_groups) & (article_namespace == 0) & (article_articleid == 0) & (new_size < 300) & new_wikitext irlike "(word|#redirect|#yönlendirme|#yönlendir|anlam ayrımı)" [18:18:44] Try: !lcase(added_lines) rlike "#(redirect|yönlendir)|anlam ayrımı") [18:18:53] ************************************ [18:19:29] The irlike is incorrect - it will treat new_wikitext as a regex [18:19:42] why? [18:19:52] because it's backwards [18:20:04] wait [18:20:04] can you explain? [18:20:05] nvm [18:20:14] I was thinking something else [18:20:26] ok [18:20:54] oh my, that is using 785 conditions! [18:20:59] irlike == case-insensitive regex match [18:20:59] what is going on here? [18:21:23] Superyetkin: The numbers do extremely jump [18:21:30] Why is "word" in your regex? [18:21:33] do you know why? [18:21:36] now it's one [18:21:50] now 25 conditions [18:22:12] yes, I see [18:22:24] but our total does not change! [18:22:25] Son 2.709 eylemde, 1.403 (%51,79) filtre 1.000 koşul sınırına erişti, ve 154 (%5,68) eylem şuanda etkin bir filtreyle eşleşti.Son 2.709 eylemde, 1.403 (%51,79) filtre 1.000 koşul sınırına erişti, ve 154 (%5,68) eylem şuanda etkin bir filtreyle eşleşti. [18:22:52] Superyetkin: Why is your regex "(word|#redirect|#yönlendirme|#yönlendir|anlam ayrımı)" [18:23:05] I don't think the word| belongs [18:23:10] remoed word now [18:23:12] thanks for that [18:23:31] Also, #yonlendir will always match when #yonlendirme does [18:23:44] So you can remove #yonlendirme too [18:24:13] done [18:24:48] and it consumes 999 conditions of the condition limit [18:25:04] I cannot understand why [18:25:09] where do these numbers come from? [18:25:50] The numbers are bugged [18:26:00] how to redecue total? [18:26:05] This filter is also problematic: https://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci/42 [18:26:19] it is not reached at all! [18:27:16] how to decrease the total conditions used? [18:27:34] Son 2.728 eylemde, 1.418 (%51,98) filtre 1.000 koşul sınırına erişti, ve 154 (%5,65) eylem şuanda etkin bir filtreyle eşleşti. [18:27:51] How could it consume 1,001 conditions of the condition limit if the condition limit is 1000, harlot? [18:28:24] hi [18:28:35] hi [18:28:35] what to use in lua to check if page exists? [18:29:04] Base-w: Who wants do this? [18:29:07] harlot, are you a developer? [18:29:27] No [18:29:31] rillke: i need to check if page exists in template [18:29:52] (in lua) [18:29:52] create a page object and then use the exists method [18:30:06] Something else is hittingyour condition limit [18:30:13] how to reduce the total number of conditions used? [18:30:13] https://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci/42 [18:30:21] Remove expensive checks [18:30:45] that is ineffective, man [18:30:50] 42 is not reached at all [18:30:52] base-w: local title = mw.title.new [18:31:14] * Base-w: local title = mw.title.new( text, namespace ) [18:31:22] harlot [18:31:27] what to do in #42? [18:31:38] rillke: thanks [18:31:46] oh my [18:31:48] or with mw.title.makeTitle [18:31:48] 52% [18:32:06] and then title:exists [18:32:38] Base-w: Note that creating a title is expensive [18:32:42] !("autoconfirmed" in user_groups) & (contains_any(added_lines, "asadasd", "asdf", "asdfg", "asdfgh", "asdfghj", "asdfghjk", "asdfghjklş", "hahhahha", "hahhah", "hahaha", ":)", ":D" )) [18:33:03] this is a template for only one page so should be ok [18:33:14] If you want to avoid this, use https://commons.wikimedia.org/wiki/Module:Link [18:33:32] added_lines rlike "asdf|hahhah|hahaha|\:[)D] [18:33:37] added_lines rlike "asdf|hahhah|hahaha|\:[)D]" [18:33:58] Most of those checks in there are redundant [18:34:18] Anything that matches asdf will by definition match asdfg [18:34:28] I agree [18:34:35] deactivating that one for now [18:36:26] what to do to lower the total: 52%? [18:39:01] Here's an expensive filter [18:39:02] https://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci/32 [18:39:29] This one too [18:39:29] https://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci/30 [18:40:16] https://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci/29 [18:41:08] This one's somewhat slow [18:41:08] https://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci/24 [18:41:40] Superyetkin: This one almost duplicates #6 [18:41:40] https://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci/22 [18:42:27] Another slow one [18:42:27] https://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci/20 [18:42:53] Really slow [18:42:54] https://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci/19 [18:43:14] how to improve #20? [18:43:20] Kinda slow [18:43:20] https://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci/13 [18:43:36] Also kinda slow [18:43:38] https://tr.wikipedia.org/wiki/%C3%96zel:%C4%B0stismarS%C3%BCzgeci/9 [18:43:39] man, #19 is from en-wiki :) [18:44:02] Nothing on enwiki should take 5.46ms to run [18:44:27] Simply put, you;ve got a ton of filters all using ~4-5ms to run [18:44:36] we have imported that one from there [18:44:47] So, ~40-50ms total [18:45:05] yes? [18:45:42] you think the total is relevant to that fact? [18:45:52] 52% [18:47:24] Yeah, your filters have some problems :\ [18:47:34] They're using a ton of server resources. [18:48:26] The problem is, I can't read Turkish, so I'm not to certain what all these filters are meant to do. [18:48:58] But, for example, why do we need a filter that does this: [18:49:00] !("autoconfirmed" in user_groups) & (contains_any(added_lines, "'''Kalın metin'''", "''Eğik metin''", "[[Sayfanın başlığı]]", "[[Sayfanın başlığı]", "[http://www.example.com adres açıklaması]", "== Başlık metni ==", "Başlık metni", "[[Dosya:Örnek.jpg]]", "[[Medya:Örnek.ogg]]", "

Matematiksel-ifadeyi-girin

", "Buraya biçimlendirilmemiş metin ekleyin", "#YÖNLENDİRME [[Hedef say [18:49:23] Looks like example code filter? [18:49:53] man, that filter is checking if the user has "randomly" put in text in the article [18:50:03] using the buttons on the edit panel [18:50:20] there are vandals doing exactly that [18:50:38] I remember deleting hundreds of articles just for that reason [18:50:54] People are creating articles solely containing test code? [18:51:04] yes [18:51:16] Try just regex-matching the text [18:51:17] we are combating serious vandalism [18:51:47] ok, will try that [18:53:00] added_lines rlike "Sayfanın başlığı|(Kalın|Eğik) metin|Örnek.(jpg|ogg)|adres açıklaması|Buraya biçimlendirilmemiş metin ekleyin|Matematiksel-ifadeyi-girin|Hedef sayfa adı|Üstü çizilecek metin ... [18:53:10] Something like that is faster than contains_any [18:56:55] yes, faster but are you sre that will lower the total? [18:57:04] we are eating up resources here: 52%!, [18:57:59] how can we monitor the filters without going through them one by one? [18:58:43] oh my [18:58:55] #6 is still high: 257 conditions ... [19:02:54] #6: 663 conditions now [19:03:06] oh, that is too costly [19:17:27] harlot, thanks for your help [19:20:49] error: unable to unlink old 'RELEASE-NOTES-1.22' (Permission denied) [19:21:05] saw this error when I ran git rebase origin/wmf/1.22wmf5 [19:21:24] Reedy: ^^ [19:23:37] aude: looks like I'm picking up one of your core changes in the deployment [19:23:46] "Update Wikibase with patches for property parser function" [19:23:49] hope that's OK [19:24:21] kaldari: should be already deployed [19:24:28] * aude thinks so [19:24:41] looks like it hasn't been deployed to wmf4 at least [19:24:46] that's where I'm picking it up [19:24:48] can't be [19:24:54] if not, please to [19:24:55] do [19:25:06] no problem :) [19:25:20] kind of bad if it's not yet there but okay [19:25:31] might be on wmf5 instead [19:25:43] ok [19:38:44] scap started [19:50:30] binasher: thanks for setting up the wikidata tables :) [19:50:43] no prob [20:00:42] rillke: i getting attempt to call method 'exists' (a boolean value). [20:00:55] i'm a very noob in lua so idk what's wrong [20:07:05] sorry then use it as a propery [20:07:22] without the invoking brackets [20:09:49] =mw.title.new( 'Module:Languages', 'Module' ).exists [20:09:52] works [20:11:45] spagewmf, superm401, ori-l: we got everything finished with our deployment except for config changes. Any chance we could piggy back them on your deployment, or do them after yours? [20:11:51] which is the same as =mw.title.new( 'Languages', 'Module' ).exists [20:12:04] kaldari, yeah, that should be fine. [20:12:11] I'll check in when we're ready. [20:12:15] Base-w:^^ [20:12:33] piggy-back or do our own afterwards? [20:12:42] rillke: works this way, thanks [20:12:58] kaldari, piggy-back. [20:13:09] Base-w: Forgot that it is a property [20:13:25] Base-w: https://www.mediawiki.org/wiki/Extension:Scribunto/Lua_reference_manual#mw.title.new [20:13:46] superm401: cool, I've merged them then [20:13:53] i see that, but i dont know lua yet :) [20:14:24] Base-w: I could remember the phrase you told me yesterday ... [20:14:37] Or was it 2 days ago [20:15:02] erm [20:15:08] * Base-w triing to recall [20:15:20] was it in this room? [20:15:48] rillke: ^ [20:16:38] I found it: "are you first day in a wiki? :)" [20:17:32] ah :D [20:17:53] Base-w: Even if its developers will never admit it, LUA has lots of common elements with VisualBasic [20:18:12] i dont know VB either [20:18:45] i know a few JavaScript and Java only [20:19:10] performing a zero-edit is easier than lua :P [20:19:22] rillke: ^ :) [20:21:30] Base-w: I disagree :P [20:21:56] lua is developing but zero-edit is using :P [20:22:18] No, no, you're confusing this. [20:22:22] *g* [20:45:11] :) [21:26:30] halp [21:26:32] Error deleting file: The file "mwstore://local-multiwrite/local-deleted/r/u/o/ruosi6mzdnkvv3vgq0cz24p4tlao35d.jpg" is in an inconsistent state within the internal storage backends [21:35:52] AaronSchulz, ^ [21:36:06] superm401: you guys still preparing for deployment? [21:36:16] kaldari, tin is giving us a hard time. [21:36:29] I still need to rebase into the wmf5 dir, but there's a permissions problem. [21:36:38] yeah, we ran into that too [21:37:01] Maybe Reedy is around now [21:38:07] Magog_the_Ogre: what wiki? [21:38:24] en.wp: Aaron|home [21:41:52] Magog_the_Ogre: try now [21:43:32] worked, thanks Aaron|home [22:02:27] hi again [22:03:34] is there a way to set not only ns0 as default namespace to search in but others too. If yes technicaly then is it allowed on wmf's wikis? [22:14:02] Base: Just do a search at Commons [22:14:14] http://commons.wikimedia.org/w/index.php?search=test+xxx&title=Special%3ASearch [22:15:57] Files (NS6), categories (NS14), galleries(NS0) [22:21:02] nice [22:21:13] Thank you [22:21:23] what variable is it? [22:23:57] rillke, ^ [22:37:58] have found $wgNamespacesToBeSearchedDefault [23:47:00] about to run scap